A Robust 3D Mesh Watermarking Approach Based on Genetic Algorithm

Abstract

In this paper, an optimized 3D watermark approach is be presented, the embedded process depends on modifying the statistical distribution radial parameter. The proposed approach consists of three Steps, the first Step depends on selecting the best vertices that will carry the watermark stream bits, these vertices called the Points of Interest (POIs). The second Step is the training process using the genetic algorithm (GA) to detect the best parameter lambda that will be used to modify the statistical distribution, this lambda grantee the optimal balance between the imperceptibility and robustness. The third Step is the embedded process by using this best lambda. The experimental results shows that the proposed approach is robust against different types of connectivity attack (like subdivision and simplifications attack) and geometrical attacks (like similarity transformation, smoothing and adding noise). The experimental results compared with the well-known method.